Relative Wages and the Market for Nursing Instructors

Transcription

1 Relative Wages and the Market for Nursing Instructors Abstract Given the reported difficulty in attracting and retaining qualified instructors, this paper examines the earnings of RNs involved in nursing instruction. Using data from the quadrennial NSSRN for 1988 through 2008, a large decrease in relative wage is found for RN instructors relative to RNs working outside of education. The paper next examines the decisions to change jobs and to moonlight. Counter to expectations, the probability of changing jobs decreased for RN educators over the period. The proportion of teaching RNs holding second jobs declines over the period while second job holding was relatively constant for non-teaching RNs. In addition, early in the sample period, RNs whose primary job was in teaching earn a wage advantage in their second job, but this wage differential goes to zero by These findings suggest that there was a market adjustment in nursing education. As the relative wage for RNs declined due to rising earnings in hospitals and other settings, nurses involved in teaching increasingly sought out and found alternative employment opportunities, leaving a different group of RNs remaining in teaching positions by the end of the period. April 2012

2 I. Introduction According to the Bureau of Labor Statistics Occupational Outlook Handbook, nursing is the largest healthcare occupation, with expected job growth greater than 22 percent over the next 10 years. 1 As a result, early research predicted RN shortages in 2020 to range from 400,000 (Auerbach, Buerhaus, and Staiger, 2007) to more than 1 million (Health Resources and Services Administration, 2004). More recent studies, however, find that due to increased entry of newer cohorts of workers these shortages may be smaller than anticipated but still substantial (Auerbach, Buerhaus, and Staiger, 2011). Explanations for these expected shortages include the combination of growing demand for nursing services, the expected retirement of an aging nursing workforce, along with an inability to increase the supply of nurses to meet this growing demand. 2 While the shortage has eased in recent years, it is expected to pick up again as the economy recovers from recession (Staiger, Auerbach, and Buerhaus, 2012). One common explanation for the lagging supply response is the constraint on nursing school capacity and enrollment. In 2010 almost 55,000 qualified applicants were turned away from BS nursing programs alone 3, and it is estimated that in 2006 as many as 150,000 were turned away from all types of programs (Aiken, 2007). The top reasons reported by nursing schools for turning away qualified applicants include insufficient faculty (71.0 percent), a lack of clinical teaching sites (55.9 percent), and not enough classroom space (50.3 percent). 4 According to a 2007 survey of nursing schools, the main reasons precluding schools from hiring additional full-time faculty include insufficient funds to hire new faculty, an unwillingness of administration to commit to additional full Aiken, 2007; Aiken and Cheung, 2008; Buerhaus, Staiger, and Auerbach, 2003, 2004; Buerhaus, Auerbach, and Staiger, 2009; Goldfarb, Goldfarb, and Long, 2008; Long, Goldfarb, and Goldfarb, 2008; Shields, 2004,Spetz and Given, 2003; Hirsch and Schumacher, American Association of Colleges of Nursing. 4 American Association of Colleges of Nursing, Annual State of the Schools,

3 time positions, an inability to recruit qualified faculty because of competition for jobs, and an unavailability of qualified applicants in the school s geographic area. 5 In order to explore the issue of nursing faculty in more detail, this paper examines the relative earnings of RNs involved in nursing instruction. Using data from the quadrennial National Sample Surveys of Registered Nurses (NSSRN) for years 1988 through 2008 it provides measures of how nurses involved in teaching have fared relative to those involved in non-academic settings. The results indicate that nurses involved in teaching experienced large decreases in their earnings relative to other nurses over this period. As the earnings of nurses working in hospitals and other nonacademic settings were bid up, the salaries for teaching positions have not kept pace, resulting in an eroding relative wage for nurse educators. Information on Job changes and nurses who hold second jobs is then used to examine the role of unmeasured ability in explaining the wage differential between instructors and non-instructors. The job change rates for nurse educators declines over the period, while it remained relative constant for other types of nurses. In addition, early in the period nurses who are instructors in their primary job receive a wage advantage in their second job relative to nurses who are not instructors in their primary job. By 2008, however, this wage advantage goes to zero. This is interpreted as a shift of relatively more able nurses involved in teaching to occupations outside of teaching as the relative wage for instructors eroded. The paper proceeds as follows. The next section provides background on the nursing instructor market. Section III discusses the NSSRN data set and provides descriptive evidence on nursing markets. Section IV examines the change in relative wages between nursing instructors and other types of nurse. Evidence on job changes and the decision to hold second jobs is examined in section V, and section VI provides conclusions. 5 American Association of Colleges of Nursing, 2007 Survey on Faculty Vacancies, Washington, DC. 3

4 II. Background As of 2008, about 20.4 percent of the RN workforce held a diploma, 45.4 percent held the associate s degree, and 34.2 percent held a bachelor s degree or higher. 6 In 2010, 2.7 percent of US trained nurses taking the NCLEX exam were from diploma programs, 29.3 percent were from Baccalaureate degree programs, and 57.9 were from associate degree programs (National Council of State Boards of Nursing, 2010). Thus, nursing instructors can be employed by colleges and universities, community colleges or technical schools, and hospital programs. RNs are also employed as instructors in LPN/LVN and nursing aide programs. Nurses involved with teaching often hold a PhD, but not always. According to the American Association of Colleges of Nursing (AACN), of the 755 vacant Baccalaureate faculty positions in the academic year, 58.8 percent required an earned doctorate, 31.8 percent required a Master s Degree, but preferred a Ph.D., 7.2 percent required a Master s Degree, and 2.3 percent required some other degree. 7 The education requirements for teaching in diploma or associate degree programs are lower. Most instructors in these programs will hold a master s degree, but some may have a Ph.D. Nurses working in education are also typically required to be involved in both classroom and clinical teaching. The AACN reports that 74.3% of Baccalaureate faculty positions required both classroom and clinical teaching, 22.5 percent involve only classroom time, and 2.1 percent only clinical teaching. 8 Thus the work of a nursing instructor includes time in the classroom and time spent with patients. Even during breaks and vacations the average nurse educator works more than 24 hours per week. During the school term nurse educators worked an average of 56 hours per week, which exceeds the average of in academics in general (National League for Nursing, 2007). 6 The Registered Nursing Population, Findings from the 2008 National Sample Survey of Registered Nurses, September 2010, US Department of Health and Human Services, Health Resources and Services Administration, 7 American Association of Colleges of Nursing, Special Survey on Vacant Faculty Positions for Academic Year , 8 Ibid. 4

5 In order to attract nurses into teaching, schools must offer enough compensation to make RNs at least indifferent between employment as a practicing nurse and that as a nursing instructor. For academics in general schools can offer better working conditions (flexible hours, summers off, rewarding and challenging work, academic freedom, etc.) to make up for lower compensation than in non-academic settings. Nursing, however, may not have as much flexibility as instructors are required to spend time in both classroom and clinical settings. Colleges and universities, however, may be reluctant to increase nursing salaries to make up for this difference. Nursing schools are expensive for colleges to operate. The provost of the Medical Education Campus of Northern Virginia Community College indicates that they lose $8,000 per year for every nurse they train. 9 Thus, colleges and universities may be unwilling or unable to increase salaries sufficiently for nursing instructors to balance the wage difference between academic and nonacademic settings. At the same time, nursing wages have risen rather rapidly and nurses are paid high wages relative to other occupations with similar skill requirements and working conditions (Hirsch and Schumacher, 2012). Due to changing technology, nursing shortages, and overall increased demand, the wages of RNs have been pushed up in both absolute and in relative terms as compared to most groups of workers in the economy. This demand increase has been largest for the higher skilled nurses who are also most qualified for teaching positions (Schumacher, 1997). There is a close association between nursing school enrollments and the earnings of RNs. Figure 1 shows an adjusted wage time series along with a relative wage index. The adjusted time series is constructed by estimating a log wage regression using data from the monthly Current Population Surveys for the years 1992 through The figure plots the coefficient on the yearly dummy variable which shows how the real wage (adjusted for inflation using the Consumer Price Index) for RNs changes relative to the base year (1992) after adjusting for measurable characteristics. Early in the period, real 9 What Works: Healing the Healthcare Staffing Shortage, Price Waterhouse Coopers, This regression includes controls for potential experience (years since finishing school) and higher order terms to the fourth power, and dummies for degree type (4), race/ethnicity (2), gender, part-time status, union membership, hospital employment, marital status (2), region (8), large metropolitan area, and year (18). 5

6 wages were falling so that an RN in 1997 earned a wage about 94 percent of an RN in 1992 with identical characteristics. After 1997, however, real wages began to rise so that by 2002 RNs were above their 1992 wage level, and an RN in 2010 earned a wage about 5 percent more than an equivalent RN in The wage index is created by pooling RNs with females with at least an associate s degree in non-healthcare occupations. A separate log-wage regression is run for each year and the coefficient on a dummy variable for RN reflects the relative wage difference between RNs and college-educated females. While the scale of this variable is different than for the adjusted timeseries, it shows the large wage advantage RNs earn relative to other occupations. It also shows a similar pattern as the adjusted wage time series of falling relative wages early in the period but increasing wages afterwards. Figure 2 shows the percentage change in enrollments in entry-level Baccalaureate nursing programs. These data show a close connection to the wage measures in Figure 1, albeit with a lag. Between 1994 and 2000 when nursing earnings were relatively flat or declining, enrollments in nursing schools declined slightly. When nursing wages began to increase in the late 1990s or early 2000s, nursing enrollments picked up rather dramatically. But enrollments could have been even larger: since 2004 at least 30,000 qualified applicants have been turned away each year from entrylevel Baccalaureate nursing programs with almost 55,000 turned away in The link between wage growth and nursing school enrollment is consistent with findings on choice of college major in general. Freeman and Hirsch (2008) find that the choice of college major is responsive to changes in employment opportunities and the knowledge content of jobs in the economy, although far less sensitive to changes in relative wages. Students have better knowledge of where the jobs are than what they pay. Thus, nursing staffing problems are likely to have worsened. If market forces outside of teaching are increasing wage and employment opportunities for nurses, but institutional constraints within academia are holding down the growth in salaries, we would expect nurses qualified to be

7 instructors to move out of teaching and into non-teaching settings. Those who remain in or continue to select into teaching would be those who have an especially strong affinity for teaching or are not sufficiently mobile to switch jobs. In the sections that follow, I examine the evidence on the relative pay for RNs in and outside of teaching. III. Data and Descriptive Evidence The Primary Data for this study are taken from the quadrennial National Sample Survey of Registered Nurses (NSSRN) for 1988 through The survey is mailed to individuals with an RN license in the U.S. every four years by the Bureau of Health Professions. The resulting public use data contain employment, education, and demographic information on the registered nurse population; including current employment setting as well as information on second jobs within nursing. The sample is restricted to those who were employees of a facility or a temporary employment agency (selfemployed were eliminated), and who were employed in nursing as their principal position. The survey asks individuals to identify both their employment setting as well as their position. The setting categories include hospital, nursing home, ambulatory care center, public or community health settings, and nursing education, among other categories. The choices within the nursing education category include LPN/LVN program, diploma program, associate degree program, baccalaureate and/or higher degree program, and other program. The options within the position categories include instructor in a school of nursing and Assistant/Associate/Full professor. I use both of these measures as two alternative measures of nursing teachers. Nursing Education is defined as those whose setting is in one of the nursing education categories, and a nursing instructor is defined as a nurse whose position is either an instructor or a professor. Table 1 displays basic demographic characteristics of nursing instructors and non-instructors. Approximately 2.2 percent of the RN sample reports their primary position as a nursing instructor. While not shown in the table, this proportion declined slightly over the period; from about 2.6% in 7

8 1988 to 1.9% in Overall, instructors tend to earn a higher hourly wage, 12 report similar hours per week, but more than 200 hours less per year. Instructors tend to be older and have more experience, but otherwise have similar characteristics to non-instructors. Real annual salary, hours of work per week, and weeks worked per year are reported in Table 2 by year and by nursing position. The data are shown for instructors, nurses involved in LPN/LVN, diploma, associate degree, and BA programs. Also shown are nurses not involved in education, staff RNs, and supervisors. 13 The real earnings of nurses increased over the period for all categories (except those involved in LPN/LVN instruction), but the increase was greater for nurses not involved with education. Though the sample sizes are small within type of education group, salary increases for these groups were lower than for those outside of teaching. The hours worked per year seem to increase slightly, and mostly between 2000 and 2008, while the number of weeks worked per year remains relatively constant. The paper next examines how accounting for differences in measurable characteristics affects this trend in earnings data. IV. Changes in the Relative Wage Associated with Nursing Instruction In order to adjust for differences in worker characteristics between nurse educators and non-educators and across time, log wage regressions are estimated separately for each survey year. Nurse educators are pooled with RNs not involved with teaching and a dummy variable on nurse educator is included. These results are presented in Table 3. In specification 1, the nurse educator category is defined as those who indicate their primary setting is in nursing education, while in specification 2 this is broken out by type of program. As an alternative, specification 3 includes dummy variables for position title. Employment in a hospital is a mutually exclusive category for the education variables in regressions 1 and 2, but not in 3 (though it is controlled for) All earnings data are reported in constant 2008 dollars. 13 Staff Nurses include Charge Nurse, Public Health Nurse, School Nurse, Staff Nurse, Team Leader, and No Position Title. Supervisor includes Administrator of Facility/Agency, Administrator of Nursing, Supervisor or Assistant Supervisor, Head Nurse or Assistant Head Nurse, and Nurse Manager. 14 Other variables included in these models are dummy variables for level of education (3 for 4 categories), experience and its square, and dummies for employment in a large metropolitan area, part-time employment, 8

9 Regression results reported in Table 3 show that, overall, nurse educators earn a wage premium relative to non-educators. Pooling the surveys over the years, those in nurse education earn a log wage about.118 points (12.5 percent) 15 more than nurses employed outside of hospitals but not in education, while a hospital employed RN earns about 3.3 percent more than an RN in education. Likewise, a nursing instructor earns about 4.0 percent higher wages than a supervisor, and about 14.9 percent higher wages than a staff nurse. Specification 2 shows that over the entire period, with the exception of teaching in an LPN program where wages are much lower, teaching RNs earn a wage at or just below those paid to hospital based RNs. Examining how these differentials change over time shows that nursing instructors experienced large relative wage losses over the period. An RN involved in nursing education in 1988 received a wage 12.2 percent higher than a hospital RN, and 32.7 percent higher than a non-hospital RN employed outside of teaching. This positive differential eroded over the period so that by 2008 a nursing educator earned about 13.3 percent less than a hospital RN and only about 2.9 percent more than RNs employed outside the hospital sector. RNs involved in teaching in a BA program saw their wage advantage over non-hospital RNs fall from almost 35 percent in 1988 to about 3.7 percent in 2008 while associate degree educators saw a decrease from 35.8 percent to about 6.5 percent. RNs teaching in a BA program in 1988 earned about 14.8 percent more than hospital RNs, but by 2008 earn about 11.1 percent less than hospital RNs. Similar results are obtained when examining the position title of the RNs. In 1988 nurse instructors earned 28.9 percent more than a staff RN and about 13.1 percent more than an RN supervisor. By 2008 these differentials fell to 4.1 and 3.1 percent respectively. race/ethnicity (3 for 4 categories), martial status (2 for 3 categories), employment by a temporary agency, gender, and region (8 for 9 categories). 15 Log wage differentials are converted to percentages by 100[exp(β)-1] where β is the log wage coefficient. 9

10 The results show that RNs involved in teaching experienced large relative wage declines over the 1988 to 2008 period. 16 While real earnings rose slightly for RN teachers (Table 2), when compared to RNs in alternative employment settings, nursing instructors earnings fell significantly. As the demand for RNs increased in non-academic settings increased, their wages were driven up. In order to keep compensation competitive, colleges and universities faced pressure to increase compensation to their RN teachers, but the evidence here suggest that RN teacher compensation did not keep pace with the rest of the RN sector. Thus, it is not surprising to find that one of the binding constraints on nursing school capacity and enrollment is insufficient faculty. V. Evidence on Job Changes and the Decision to Hold a Second Job The results above showed that when compared to nurses employed in other sectors, nursing educators experienced substantial declines in relative wages. Given that RNs, once educated, are mobile between teaching and practicing nursing we would expect that as relative wages decreased, teaching nurses would move out of teaching positions and into other positions. Many of the explanations reported by nursing schools for the difficulty in filling vacant faculty positions deal with the inability to compete with earnings outside of teaching. 17 This movement, however, is not likely to have been random. We would predict that the most able or most mobile RNs would be most likely to leave first. This section first examines job change rates among RNs and then explores the decision to hold a second job and the earnings on that second job. Beginning in 1992, the NSSRN survey asks if the individual is in the same job as they were a year ago and, if not, asks them about the setting and position of their previous job. This allows an estimate of the rate of job change by RN type. In surveys prior to 1992, RNs were asked if they were 16 Restricting the RN comparison group to include only RNs with at least a Baccalaureate or at least a Master s degree results in the coefficients falling by 1 to 2 percentage points, but the trends remain the same. 17 American Association of Colleges of Nursing, 2007 Survey on Faculty Vacancies, Washington, DC. 10

11 in the same position as a year ago but they are not asked about their previous setting or position. Thus, we cannot analyze job changes in 1988 with these data. As relative wages fell in the teaching sector, one would predict that, all things equal, exit rates for instructors should increase over the years. That is, as the opportunity cost of remaining in teaching increased we should see more movement out of teaching and into other sectors of nursing. The first two columns of Table 4 display job change rates for RNs by job setting status in the previous year of the survey. Overall, both teaching and non-teaching RNs had about an 11 percent job change rate. Job change rates by year, however, show that while the rate for non-teachers was relatively constant, there was a decline in the job change rate for RN educators; from about 14 percent in 1992 to 8 percent in The final column of Table 4 shows the results of a probit on the decision to change jobs. Shown is the marginal effect (and standard error) for RN educators relative to non-educators on the job change decision. In 1992 RN educators are about 5 percentage points more likely to change jobs than non educators. This falls to about 3 percentage points in 1996 and is marginally significant. By 2000 there is no longer a significant difference in job change rates. So as the relative wage advantage for teaching nurses eroded over the period, they were initially more likely to change jobs than other nurses. But by 2008, teaching RNs had a slightly lower job change rate than those outside of education. Thus, theory predicts that as relative wages decline one should see an increase in job mobility, yet the results in Table 4 show the opposite. One explanation to this puzzle is that the relative ability of nursing instructors declined over the period. If everyone was the same ability (conditional on the non-wage X s) then as earnings in the teaching sector declined, we should see higher exits among teachers. However if the most able instructors are most likely to exit teaching, then the lower exit rates in later years will reflect low unmeasured quality (observable to employers but not to the researcher) and higher exit rates in the early years reflect higher unmeasured quality. 18 Both groups had an unusually high job change rate in

12 As an alternative to changing jobs, nurses may attempt to supplement earnings on their primary job through moonlighting. The literature on the decision to hold a second job emphasizes the role of the constraint on hours worked on the primary job (Shishko and Rostker, 1976; Krishnan, 1990; Conway and Kimmel, 1998). Consistent with this theory, Paxson and Sicherman (1996) find evidence that workers who want to work more hours in their primary job are more likely to take second jobs, movements into (out of) second jobs results in large increases (decreases) in total hours, and that workers do not decrease their main job hours when they take second jobs. Other reasons for taking second jobs focus on the possibility that different jobs may not be perfect substitutes (Conway and Kimmel, 1998), or workers may want to learn about new occupations or gain training (Paxson and Sicherman, 1996). Workers in salaried jobs are more likely to be constrained than hourly workers as increases in hours do not directly increase earnings. Thus instructors cannot readily increase earnings as they will typically be salaried, while hospital-based nurses are more likely to be paid hourly. The NSSRN asks nurses if they hold a second job in nursing. 19 If the individual answers yes, they are then asked further information about that job, including position, hours worked, and earnings. 20 Table 5 displays characteristics of RNs with and without second jobs. About 15 percent of the samples hold a second job for pay. Paxson and Sicherman (1996), using data from the Panel Study of Income Dynamics for the years , find that, economy-wide, in any given year about 21 percent of males and 12 percent of females hold second jobs. Hipple (2010) estimates point- 19 In surveys prior to 2004, the question is stated as: Do you hold more than one position in nursing for pay? The 2004 survey states the question as: Aside from the principal nursing position you just described, do you hold any other positions in nursing for pay? The 2008 survey asks: Aside from the principal nursing position you just described, did you hold any other positions in nursing for pay on March 10, 2008? That is, the 2008 survey asks if the individual holds other positions in nursing at the time of the survey, while earlier surveys ask the question so that it could be interpreted in a broader way. If the individual knew they would be working in another job later in the year, for example, they might say no to the question as it was phrased in 2008 but yes in years prior to that. Therefore, if nurses involved with education frequently work in a second job over the summer, they might say yes to the question in the early years but no to the 2008 question. 20 The 2004 survey changed the way hours worked was collected. Previously they mirrored the questions on the primary job and asked individuals for the typical hours worked per week and the number of weeks worked per year. Beginning in 2004, however, the survey asks workers to choose among a discrete number of annual hours categories. Thus, the average hours worked over time may not be consistent in this survey year as compared to others. The 2008 survey reverted back to the method used in the pre 2004 surveys. 12

13 in-time job holding rates from CPS data to be around 5 or 6 percent for males and females. RN Second job holders tend to be younger, have less experience, are less likely to work in a hospital in their primary job, and less likely to be married. Finally, if a worker s primary job is in education they are more likely to hold a second job. Table 6 shows the proportion with a second job, and the primary and secondary wage of RNs with second jobs by year. Among RNs whose primary job was not in teaching, the proportion of second job holders increased between 1988 and 1992, but then stayed at about 15 percent until 2008 when it fell to about 12 percent. Among RN teachers, the rate of second job holding is higher, but there is a more pronounced decline after Part of the decline between 2004 and 2008 could be due to the way the 2008 question was worded, 21 but there appears to be a slight decline in the proportion of teaching RNs who hold second jobs. The hourly wage tends to be higher in the second job, presumably since these jobs are less likely to have non-wage benefits. Mirroring the results for the whole sample, the primary wage for teachers remains relatively constant, while the wage for nonteachers increases. The second wage is higher for teachers than non-teachers, but the difference declines over the period. The distribution of second jobs is quite similar for those whose primary job is within education and those whose primary job is outside of education. Among those who hold second jobs and whose primary job is in nursing education, about 14 percent of these jobs are in teaching and about 47 percent are in hospitals. Among second job holders whose primary job is not in nursing education, about 12 percent of these second jobs are in teaching and 43 percent are in hospitals. The final column of Table 6 shows the marginal effect on a nursing education dummy in probit regressions on the decision to hold a second job. 22 Similar to what the raw probabilities show, nurse educators are more likely to moonlight than their non-educator counterparts, but this effect has diminished over time. An RN whose primary job in 1988 was in education, was 21 percentage points 21 See footnote Other variables included in the model were years of experience and its square, and dummy variables for degree type, large metropolitan area, race/ethnicity, marital status, and gender. 13

14 more likely to work a second job than a nurse not in education. By 2004, this fell to 12 percentage points. 23 One explanation for the higher level of second job holding is that salaried nurse educators are more likely to be constrained in increasing their annual earnings through increased hours of work than are RNs paid by the hour and have a higher demand for secondary employment. The decline in the rate of second job holding can be explained by market adjustment. As relative wages declined over the period, nurse-educators who most desired additional hours and earnings may have found permanent positions outside of education, especially so over the 2000 to 2008 period when nursing shortages were common. Nursing educators who remained as educators would then be those least likely to desire additional hours and earnings from a second job. The decline in the probability of holding a second job closely mirrors the decline in the probability of changing jobs for nursing educators, and is consistent with a market adjustment having occurred during the earlier years. As the relative wage for RNs declined due to rising earnings in hospitals and other settings, nurses involved in teaching increasingly sought out and found alternative employment opportunities, leaving a different group of RNs remaining in teaching positions by the end of the period. To examine this adjustment further, the paper now turns to examining how the primary job setting affects the secondary wage. Consider the following wage equation: Where lnw si is the log wage in the second job for worker i, X is a vector of worker characteristics and β their coefficients, ED si is a dummy variable for education in the secondary job and ED pi is a dummy variable for education in the primary job. The coefficient δ 2 shows the effect of the worker s primary job on their secondary wage, holding constant the worker s secondary job setting and other measureable characteristics. It is possible that education in the primary job directly affects the worker s secondary wage if teaching results in higher second job productivity. For example, teaching could give a nurse access to new skills or techniques that are directly used in their second job. More 23 Similar results are obtained when we use nurse instructor as the nurse education variable, or if we include only those RNs involved in teaching in baccalaureate programs. 14

15 likely, however, is that this variable will capture unmeasured worker attributes correlated with the secondary wage. If nursing educators tend to have higher unmeasured skill, then these higher skills are likely to be rewarded in both the primary and secondary job and would result in a positive value for δ 2. Table 7 provides log wage regression results for RNs on the primary and secondary job. The first two columns use the log wage on the primary job as the dependent variable, but the first column includes the whole sample, while the second column restricts the sample to be those with second jobs. The coefficients on the variables are quite similar. So while Table 5 suggests that the X s are similar for those with and without second jobs, Table 7 suggests that the β s are also quite similar. The final column of Table 7 uses the log wage in the second job as the dependant variable. In general, the magnitude of the coefficients are lower than for the regression on the primary wage. The coefficient on the primary job in teaching dummy suggests that RNs whose primary job is in teaching earn about 8 percent more in the second job. This suggests that either teaching provides workers with some direct skill that is valued in the second job, or that RN educators, on average, have higher unmeasured ability than non educators. Table 8 displays the effect of teaching in the primary job on the second job wage over time. A separate regression is run for each year where the dependent variable is the log of the wage in the second job. The first column uses the job setting in the primary job to define working in education while the second column uses the primary position. Both definitions tell the same story. In 1988 an RN whose primary job was in education earned a large premium in her second job relative to those whose primary job was outside of teaching. This premium increases in 1992, but then falls rather dramatically so that by 2004 there is no longer a significant difference in the secondary wage. This implies that either the direct effects of teaching on the productivity of the second job diminished, or that the unmeasured worker attributes positively correlated with wages diminished for teaching RNs 15

16 relative to non-teaching RNs. 24 That is, as the relative wage on the primary job fell (as shown in Table 3), job change rates were higher for teaching RNs than non-teaching RNs (Table 4) consistent with a movement out of teaching and into practice. This movement, however, was most likely for teaching nurses with the highest unmeasured ability. The decline in the coefficient on the dummy variable for teaching in the primary job in Table 8 is consistent with this relative decline in unmeasured ability. VI. Conclusions Reports of chronic nursing shortages and predictions of larger shortages looming has caused policy concern. At the same time, nursing schools have reported difficulties attracting and retaining instructors and, therefore, have had to turn away thousands of qualified applicants. The results here document that nursing instructors experienced large decreases in relative wages when compared to RNs working in non-educational settings. In 1988 a nursing instructor earned about 30 percent more than a staff RN, but by 2008 this differential fell to about 4 percent. Information on job changes documents that teaching RNs were significantly more likely to report job changes in the 1992 and 1996 surveys, but equally likely afterwards. Similarly, nurse educators are more likely to hold second jobs but there was a decline in relative probability over the period. Finally, examining the earnings of workers in second jobs shows a decrease in the wage premium RNs educators (in their primary job) receive in their second job. This is interpreted as a decrease in the unmeasured ability of nursing educators relative to non-educators. As nursing wages in hospital and other settings were bid up over this period, the salary of nursing instructors did not keep pace making it difficult for nursing schools to attract qualified instructors. This led to a decrease in the relative skill level of nursing instructors as the most able or most mobile nurse educators were more likely to leave for better opportunities outside of teaching. If policy makers hope to resolve 24 Some nurse educators may work part-time in clinical positions to keep their skills current and, thus convenience and control over hours may be more relevant than wages. This, however, would lead to lower second job wages, and unless these preferences for non-wage amenities increased for nurse administrators, this would not affect the trend in the wage differential. 16

17 expected future nursing shortages by increasing the output of nursing schools, they first need to establish ways to increase the compensation of nursing instructors relative to what these nurses could make in alternative employment settings. 17

18 References Aiken, Linda H. U.S. Nurse Labor Market Dynamics Are Key to Global Nurse Sufficiency, Health Services Research, Volume 42 (3) Part II, June 2007, Aiken, Linda H., and Robyn Cheung Nurse workforce challenges in the United States: Implications for policy. OECD Health Working Papers, No. 35. American Association of Colleges of Nursing Advancing Higher Education in Nursing: 2009 Annual Report, Washington D.C. < Auerbach, David I., Peter I. Buerhaus, and Douglas O. Staiger Better Late than Never: Workforce Supply Implications of Later Entry into Nursing, Health Affairs, Volume 26(1), January/February: Registered Nurse Supply Grows Faster Than Projected Amid Surge In New Entrants Ages 23 26, Health Affairs, 30(12) December: Buerhaus, Peter I., David I. Auerbach, and Douglas O. Staiger The recent surge in nurse employment: Causes and implications. Health Affairs 28:4: w657-w668. Buerhaus, Peter I, Douglas O. Staiger, and David I. Auerbach, Is the Current Shortage of Hospital Nurses Ending? Health Affairs, Volume 22(6) November/December 2003: New Signs of a Strengthening U.S. Labor Market? Health Affairs 23 Web Exclusive, Freeman, James A., and Barry T. Hirsch, College Majors and the Knowledge Content of Jobs, Economics of Education Review, Vol. 27, No. 5, October 2008, pp Goldfarb, Marsha G., Robert S. Goldfarb, and Mark C. Long Making sense of competing nursing shortage concepts. Policy, Politics, & Nursing Practice, 9:3, Hipple, Steven F. Multiple Job Holding During the 2000s, Monthly Labor Review, Vol 133, No. 6, July 2010, pp Hirsch, Barry T., and Edward J. Schumacher Underpaid or Overpaid? Wage Analysis for Nurses Using Job and Worker Attributes, Southern Economic Journal, forthcoming.. Classic or New Monopsony? Searching for Evidence in Nursing Labor Markets, Journal of Health Economics, Vol. 24 (5), September 2005, pp Health Resources and Services Administration. What s behind HRSA s Projected Supply, Demand, and Shortages of Registered Nurses? September Long, Mark C., Marsha G. Goldfarb, and Robert S. Goldfarb Explanations for persistent nursing shortages. Forum for Health Economics & Policy, 11:2, article 10 < 18

19 National Council of State Boards of Nursing Number of candidates taking NCLEX examination and percent passing, by type of candidate, 2009, < National League for Nursing, More Findings from the NLN/Carnegie National Survey: How Nurse Educators Spend Their Time, Nursing Education Perspectives, Volume 28, No. 4, September/October 2007, pp Schumacher, Edward J. The Earnings and Employment of Nurses in an Era of Cost Containment, Industrial and Labor Relations Review, Volume 55 (1) October 2001, pp Schumacher, Edward J. Relative Wages and the Returns to Education in the Labor Market for Registered Nurses, Research in Labor Economics, Volume 16, 1997, pp Shields, Michael A Addressing nurse shortages: What can policy makers learn from the econometric evidence on nurse labour supply? Economic Journal 114, F464-F498. Spetz, Joanne, and Ruth Given The future of the nurse shortage: Will wage increases close the gap? Health Affairs 22:6, Staiger, Douglas O., David I. Auerbach, and Peter I Buerhaus Registered Nurse Labor Supply and the Recession Are We in a Bubble? New England Journal of Medicine, March 21, 2012 ( /NEJMp ). 19

20 Table 1: Characteristics of Nursing Instructors vs. Non-Instructors Instructor Non-Instructor Real wage (13.79) (10.95) Hours Worked per Year (648.67) (601.79) Age (9.63) (10.61) RN experience (10.35) (11.33) Large Metro (0.444) (.433) Part Time (0.443) (.445) Black (0.198) (.184) Female (0.165) (.235) Currently Married (0.437) (.451) Sample size 3, ,542 Data are from the NSSRN for the years 1988, 1992, 1996, 2000, 2004, and Instructors are those RNs who indicate their primary position is either as an instructor in a school of nursing or a professor (full, associate or assistant) of nursing. Real wage is in 2008 dollars. 20

21 Table 2: Annual Earnings, Hours of Work, and Sample Size by Year by Nursing Specialty Year Instructor LPN Program All years Real Earnings Annual Hours Weeks/Year N 1988 Real Earnings Annual Hours Weeks/Year N 1992 Real Earnings Annual Hours Weeks/Year N 1996 Real Earnings Annual Hours Weeks/Year N 2000 Real Earnings Annual Hours Weeks/Year N 2004 Real Earnings Annual Hours Weeks/Year N 2008 Real Earnings Annual Hours Weeks/Year N $48, ,293 $44, $46, $ $51, $52, $53, $47, $49, $43, $45, $50, $48, $47, Diploma Program $54, $48, $54, $52, $56, $60, $53, AD Program $51, ,118 $45, $47, $48, $50, $53, $55, BA Program $58, ,630 $53, $57, $54, $58, $61, $64, Non- Educational $51, ,542 $42, ,742 $48, ,961 $49, ,909 $51, ,579 $57, ,863 $59, ,488 Staff RN $46, ,506 $39, ,044 $44, ,433 $44, ,106 $45, ,587 $55, ,368 $54, ,968 Supervisor $63, ,656 $53, ,344 $59, ,905 $59, ,534 $64, ,540 $69, ,886 $74, ,447 Data are from the NSSRN. Instructors are those RNs who indicate their primary position is either as an instructor in a school of nursing or a professor (full, associate or assistant) of nursing. LPN program, Diploma Program, AD program, BA Program, are based on they primary setting question in the survey. Staff RN and Supervisor are based on the primary position question. Staff RNs include Charge Nurse, Public Health Nurse, School Nurse, Staff Nurse, and Team Leader. Supervisors include Administrator of Facility/Agency, Administrator of Nursing, Supervisor or Assistant Supervisor, Head Nurse or Assistant Head Nurse, and Nurse Manager. The 2008 NSSRN does not include weeks per year in the public use file. Real earnings are in 2008 dollars. 21

22 Table 3: Log Wage Coefficients on the Returns to Nursing Education Specification 1 Specification 2 Specification 3 Year Hospital Nursing Education Hospital LPN Program Diploma Program Associate Program BA Program Instructor Supervisor Staff RN All years.151 (.002).118 (.005).151 (.002).071 (.016).125 (.018).154 (.010).125 (.008).057 (.006).017 (.003) (.002) (.004).283 (.013).161 (.004).266 (.034).157 (.040).306 (.023).299 (.019).140 (.012).007 (.007) (.006) (.004).182 (.013).188 (.004).124 (.036).123 (.057).227 (.023).181 (.019).124 (.012).055 (.007) (.006) (.004).137 (.013).142 (.004).093 (.039).084 (.055).192 (.024).115 (.018).061 (.012).049 (.007) (.006) (.004).110 (.013).128 (.004).135 (.039).123 (.059).132 (.026).122 (019).101 (.015).047 (.006) (.005) (.005).076 (.015).129 (.005) (.042).102 (.052).123 (.024).100 (.023).041 (.018).009 (.008) (.006) (.004).029 (.010).154 (.004) (.036).064 (.031).063 (.019).036 (.018) (.015) (.007) (.005) Data are from the NSSRN. Instructors are those RNs who indicate their primary position is either as an instructor in a school of nursing or a professor (full, associate or assistant) of nursing. LPN program, Diploma Program, AD program, BA Program, are based on they primary setting question in the survey. Staff RN and Supervisor are based on the primary position question. Supervisors include administration, head nurse, and supervisor categories. Shown are log wage coefficients on nursing educator dummy variables. Specification 1 includes a dummy variable for nurses whose primary setting of employment is in nursing education. Specification 2 breaks this into separate categories for type of program. Specification 3 includes dummies for position title. Other title is the omitted group. Hospital is a mutually exclusive category in regressions 1 and 2, but not in 3 (though it is controlled for). Other variables included in these models are dummy variables for level of education (3 for 4 categories), experience and its square, and dummies for employment in a large metropolitan area, part-time employment, race/ethnicity (3 for 4 categories), martial status (2 for 3 categories), employment by a temporary agency, gender, and region (8 for 9 categories). 22

23 Table 4: Job Changes among RNs Proportion who Change Jobs Non Education in Education in Previous Job Change Probit Previous year Year All Years (.006) (.017) (.016) (.018) (.011) (.010) Data are from the NSSRN. Surveys prior to 1992 did ask about employment setting in the previous year, so while job changes can be identified, we cannot identify who was in education the previous year. The probit results are the marginal effects and the associated standard error for those RNs initially in education. Other variables included in the model are experience and it s square, and dummy variables for large metropolitan area, race/ethnicity (3) marital status (2), gender, and region (8). 23

24 Table 5: Characteristics of RNs With and Without Second Jobs Second Job =0 Second Job =1 Real Wage (in primary job) (10.94) (11.53) Age (10.77) (9.57) Experience (11.50) (10.21) Female.944 (.230).931 (.254) Hospital (in primary job).626 (.484).583 (.493) Nursing Education (in primary job).020 (.141).055 (.229) Married.723 (.448).686 (.229) Sample Size 126,961 22,263 Source: data are from the NSSRN for 1988, 1992, 1996, 2000, 2004, and

25 Table 6: Real Wage of RNs with a Second Job Primary job in Teaching Primary Job not in Teaching Proportion With a Second Job Real Wage in primary Real Wage in Second Proportion With a Second Real Wage in primary Job Real Wage in Second Second Job Probit Job Job Job Job (.008) All Years (.022) (.023) (.022) (.021) (.017) (.014) Source: data are from the NSSRN for 1988, 1992, 1996, 2000, 2004, and Real wages are in 2008 dollars. The 2004 survey changed the way hours worked was collected. Previously they mirrored the questions on the primary job and asked individuals for the typical hours worked per week and the number of weeks worked per year. In 2004, however, the survey asks workers to choose among a discrete number of annual hours categories on for hours on the second job. Thus, the average hours worked over time may not be consistent in this survey year as compared to others. The 2008 survey returns to asking typical hours per week and weeks per year on the second job. The second job probit column shows the marginal effect (and associated standard error) for RNs involved in education on the decision to hold a second job. Other variables included in the model are experience and its square, and dummies for degree type (3), metropolitan area, race/ethnicity (3), marital status, (2), gender, and region (8). 25

26 Table 7: Log Wage Regression Results for Primary and Second Job Dependent Variable Real Wage in Primary Job Real Wage in Primary Job Real Wage in Second Job Second job in Hospital (.005).114 (.010) Second Job in Teaching (.008).016 (.016) Primary Job in Hospital.151 (.002).137 (.005).026 (.010) Primary Job in Teaching.123 (.005).137 (.005).082 (.022) Associate Degree.012 (.002).008 (.007).001 (.014) BS Degree.060 (.002).048 (.007).049 (.014) Graduate Degree.143 (.002).132 (.007).099 (.013) Experience.016 (.000).014 (.001).005 (.002) Experience Squared/ (.006) (.002) (.004) Metro.107 (.002).119 (.006).100 (.011) Hispanic.035 (.006).025 (.017).072 (.032) Black.057 (.005).064 (.012).058 (.023) Other.039 (.004).033 (.012).049 (.024) Married (.003) (.008) (.016) Female (.004) (.010) (.019) Sample Size 144,744 17,767 17,767 Source: data are from the NSSRN for 1988, 1992, 1996, 2000, 2004, and Column 1 uses the log of primary wage as the dependent variable and includes the full sample. Column 2 restricts the sample to those with a second job but uses the wage on the primary job as the dependent variable and column 3 uses the wage on the second job as the dependent variable. Other variables included in the model are dummies for previously married, region (8), and year (4). 26

27 Table 8: Second Job Log Wage Coefficients by Year Primary Job in Education Primary Job as Instructor (.061).078 (.046) (.055).114 (.045) (.063).029 (.053) (.056) (.050) (.056) (.052) (.040) (.023) Source: data are from the NSSRN. The dependent variable is the real wage in the secondary job. Shown are coefficients from two separate regressions for each year. Other variables included in the regressions are years of experience and its square, dummies for degree type (3), large metropolitan area, race/ethnicity (3), marital status (2), gender, region (8), hospital employment in the primary job, hospital employment in the second job, and second job as an instructor. 27